Re: FYI: Status of gdb usage of gerrit

[Carlos O'Donell <carlos at redhat dot com>]

On 11/28/19 1:47 AM, Florian Weimer wrote:
> * Carlos O'Donell:
> 
>> * A list of things to review (patch tracking/comment tracking)
>>   - Critically important for me when I set aside time I need to jump right in.
>>   - Nothing provided by email.
> 
> I don't know how to get this into Thunderbird.  I would expect that
> you could get some basic patch tracking by using the labels (via the
> 1/2/3 keys).
> 
> With an open storage format such as Maildir, it should be too hard to
> script something that gives you a list of patches which have no
> follow-ups.

I want patch authors to update the status of the patches in the patch tracker.
I don't want to have to do that work. They should make sure their patch is clean
in our system of record and then when I go to said system I can quickly review
patch-after-patch.

Moving this work to the reviewer side is going to slow down precious review,
and it doesn't scale.

In addition the above scripts would just be inferior to patchwork which is
already doing what you mention.

Does that make sense?

I use mutt/thunderbird, and I don't mind switching to mutt just for patch
review work.

>> * A way to do side-by-side diffs and version-to-version diffs (diffuse
>> + git range-diff)
>>   - git difftool --tool=diffuse/kompare/meld
>>   - git range-diff
>>   - Needs automation to pull down the patches and create branches/commits.
> 
> That is really a local thing.  Not sure how to get patches out of the
> Thunderbird mail store, though.

As I mentioned I could just switch to mutt-only for patch review, I'm
comfortable with mutt (10+ years of use).

Even *today* I use pwclient to get patches from the mailing list. I don't
use thunderbird to extract the patches. This is why even an unmaintained
patchwork has value to *me*.

I get a bit angry with overseers when pwclient is blocked because it uses
xmlrpc and that has some rate limiting that I run into every once in a
while when I pull lots of patches.

>> * A way to hook in CI/CD
>>   - Need a way to know if the patch still applies (scripts will help)
>>   - Need a way to know basic x86_64 still works.
> 
> I view this as completely unrelated.  Anyone can pipe posted patches
> into some CI tool and post the results *today*.

Correct, but the more that is ready ahead of time the better. If the event
is triggered by my review then it's too late.

-- 
Cheers,
Carlos.